System and method for identification of fingerprints

Abstract

A method for automated fingerprint identification and special purpose programmed computer for carrying out the method are disclosed. The method and system comprise identifying a scar region in a set of pixels in a fingerprint image by using a second order Gaussian derivative filter parameterized using standard deviation, wherein the scar region has a width greater that the width of a regular ridge or a valley of separation between ridges; wherein a boundary, intensity, and flow direction of each pixel is calculated; each pixel being connected to a next adjacent pixel using flow co-ordinates and flow directions; wherein the Gaussian derivative filter has an output and the output is thresholded to create an output mask with value=1 at the scar regions and value=0 outside the scar regions; and changing the intensity of pixels in the identified scar region with the intensity of the nearest boundary pixel until the intensities of all the pixels in the scar region are changed to the intensity of their flow direction identified boundary pixels.

What is claimed is: 
     
       1. A method for automated fingerprint identification comprising:
 identifying a scar region in a set of pixels in a fingerprint image by using a second order Gaussian derivative filter parameterized using standard deviation, wherein the scar region has a width greater that the width of a regular ridge or a valley of separation between ridges; wherein a boundary, intensity, and flow direction of each pixel is calculated; each pixel being connected to a next adjacent pixel using flow co-ordinates and flow directions; wherein the Gaussian derivative filter has an output and the output is thresholded to create an output mask with value=1 at the scar regions and value=0 outside the scar regions; and 
 changing the intensity of pixels in the identified scar region with the intensity of the nearest boundary pixel until the intensities of all the pixels in the scar region are changed to the intensity of their flow direction identified boundary pixels. 
 
     
     
       2. The method of  claim 1  comprising decomposing the fingerprint image into directional images by means of a directional filter bank. 
     
     
       3. The method of  claim 2  wherein the directional images are analyzed to calculate local energy of features present in a specific direction, selecting s specific direction with a large energy component, and calculating the flow direction associated with each pixel. 
     
     
       4. The method of  claim 1  comprising creating a binary mask image by applying the second order Gaussian derivative filter to provide high response to scar regions and low response to the corresponding regular ridge and valley region. 
     
     
       5. The method of  claim 4  comprising using the binary mask image to identify the boundary curve of the scar region. 
     
     
       6. The method of  claim 4  comprising filling the scar region with boundary pixels and calculating non-scar neighboring pixels using flow directions. 
     
     
       7. The method of  claim 6  comprising moving the boundary inward as boundary scars are filled until all scar pixels are filled.